An audit of the accuracy of medication information in electronic medical discharge summaries linked to an electronic prescribing system

Abstract

Background:

Poor communication of medication information to general practitioners when patients are discharged from hospital is a widely recognised problem. There has been little research exploring the accuracy of medication information in electronic discharge summaries (EDS) linked to hospital e-prescribing systems.

Objective:

To evaluate the accuracy of medication lists and medication change information in EDS produced using an integrated e-prescribing and EDS system (where EDS discharge medication lists were imported from discharge e-prescription records, medication change information was manually entered, and medications were dispensed from paper copies of the patients' e-prescriptions).

Method:

Retrospective audit of EDSs for a random sample, representative of adult patients (n = 87) discharged from a major teaching hospital. EDS medication lists were compared to pharmacist-verified paper discharge prescriptions (considered to be the most accurate discharge medication list) to identify discrepancies. EDS medication change information was compared to medication changes identified by comparing pharmacist-verified “Medication History on Admission” forms with pharmacist-verified paper discharge prescriptions.

Results:

There were 85/87 (98%) EDSs that included a discharge medication list. Of these, 50/85 (59%) contained one or more medication list discrepancies (median 1, range 0–15). The most common discrepancy was omission of medication (58%); 84/131 (64%) discrepancies were considered clinically significant (risk of adverse outcome); 162/351 (46%) clinically significant medication changes were stated in the EDS; and 153/351 (44%) changes were both stated and included a reason.

Conclusion:

EDS discrepancies were common despite integration with e-prescribing. Eliminating paper prescriptions, enhancing e-prescribing/EDS functionality and involving pharmacists in EDS preparation may reduce discrepancies.

Keywords

patient discharge summaries continuity of patient care electronic health records electronic prescribing hospitals medication errors computerised physician order entry

Introduction

Poor communication of medication information to general practitioners (GPs) when patients are discharged from hospital is a widely recognised and long-standing problem (Bergkvist et al., 2009; Garrett and McCormack, 2014; Kripalani et al., 2007). In Australia and internationally, it has been reported that medical discharge summaries often contain inaccurate medication lists and changes made to medication regimens are often not documented or explained (Belleli et al., 2013; Cornu et al., 2012; Hammad et al., 2014; Kripalani et al., 2007; Shakib et al., 2009; Tan et al., 2014). Failure to adequately communicate discharge medication information to the GP and/or the next care provider may increase the risk of post-discharge prescribing errors, failure to implement appropriate drug monitoring and unplanned re-hospitalisation (Coleman, 2003; Forster et al., 2003; Forster et al., 2005; Kripalani et al., 2007; Stowasser et al., 2002; Witherington et al., 2008). In a prospective cohort study at a Canadian tertiary hospital, 59% of preventable and ameliorable post-discharge adverse events were attributed to poor peri-discharge communication (Forster et al., 2003).

Many strategies to improve the quality of medication information in discharge summaries have been evaluated, including guidelines and performance indicators, staff education and discharge summary templates with specific sections for medication information (Hammad et al., 2014; Kripalani et al., 2007). Despite these changes, poor quality of discharge medication information persists (Hammad et al., 2014; Roughead et al., 2016). Electronically generated discharge summaries (EDS) are increasingly used in place of handwritten summaries. The impact of EDS on the accuracy of medication information is variable and depends on how the information is entered into the discharge summary. When medication lists and information about medication changes are manually typed, the EDS may be no better than handwritten discharge summaries (Callen et al., 2010; Mills et al., 2016).

EDS that are integrated with an electronic discharge prescribing (e-prescribing) system, enabling the EDS medication list to be imported from the e-prescription record, should be more accurate. However, medication list discrepancies could still occur if, for example, medication changes are made in the e-prescribing system after the EDS has been signed-off, or if the e-prescription is printed for dispensing purposes (which is a usual practice in Australia, because paper prescriptions are required by law for dispensing and payment under the government’s Pharmaceutical Benefits Scheme). When the e-prescription is printed, amendments such as a medication cessation, addition or dose alteration, may be made on the paper copy. These amendments will not be reflected in the e-prescription record if the change is not also made in the electronic system. For medication changes, unless the e-prescribing system can capture these, and the reasons they are made, and auto-populate this information in the EDS, such information still needs to be manually entered in the EDS, creating risk of incorrect or incomplete information.

The aim of this study was to evaluate the accuracy of medication lists and medication change information included in the EDS produced using an integrated e-prescribing and EDS system, in which medication lists were imported from the discharge e-prescribing record and information on medication changes was manually entered. A secondary aim was to compare the accuracy of EDS medication lists to patient medication lists (PMLs; patient-friendly medication summaries designed to supplement verbal counselling provided by the pharmacist) prepared by hospital pharmacists.

Method

A retrospective audit of a random sample of EDS and PML for adult patients discharged from inpatient wards at a major teaching hospital in Melbourne, Australia, between April and June 2014 was conducted.

Electronic information and prescribing systems used in the study hospital

The hospital had been using an integrated e-prescribing and EDS software system (Cerner Millennium) since 2011. All prescribing was done by hospital doctors within the Cerner system. Inpatient medication charts were electronic (paperless), while outpatient and discharge prescriptions were prepared electronically and then printed and signed by hospital doctors.

Discharge medication information process at the study hospital

The process for producing discharge prescriptions, EDS and PML at the hospital was as follows:

A hospital doctor, usually a junior doctor (intern or resident), prepared a discharge prescription using the e-prescribing system, then printed and signed a paper copy. Hospital policy was that all medications to be taken after discharge should be included on the prescription, regardless of whether all medications needed to be supplied, to ensure an accurate electronic medication record.

A hospital pharmacist reviewed the paper discharge prescription and performed medication reconciliation by comparing the paper prescription with the patient’s electronic inpatient medication chart and their pre-admission medication history (which had been recorded and verified by a pharmacist upon admission to hospital). The purpose of this was to identify unintended discrepancies on the discharge prescription (e.g. omitted medications, unnecessary medications, dose errors, dose form errors).

The hospital pharmacist discussed discrepancies with the hospital doctor, and amendments to the discharge prescription were agreed:

For amendments that did not require a new paper prescription (e.g. cessation of a medication, addition of a medication that did not need to be dispensed because the patient already had a supply at home, change of dosage/directions), the pharmacist annotated the amendment on the paper prescription (and requested endorsement of the amendments by the doctor if required for legal or pharmaceutical benefit reimbursement requirements). The hospital doctor was expected to also make the amendment in the e-prescription record, but there was no process to check that this was done.

For other amendments, the pharmacist requested a new paper prescription be printed and signed by the hospital doctor after the e-prescription record had been modified accordingly.

As part of the medication reconciliation process, the hospital pharmacist annotated on the printed copy of the prescription the change status for each medication (e.g. new, dose reduced) and noted any pre-admission medications that had been ceased, to facilitate production of the PML which includes a summary of medication changes.

Using the paper prescription, the discharge medications were dispensed and the PML was prepared by the hospital’s pharmacy department (this involved manual entry of medication data into dispensing software to produce dispensing labels and the PML). Copies of the processed paper prescription and the PML were stored electronically by the hospital’s pharmacy department.

The hospital doctor, again usually a junior doctor, prepared the EDS:

The electronic record of the discharge prescription was imported into the EDS, by clicking on a link within the EDS.

Information about medication changes and reasons for changes was manually entered into the EDS.

(Note: Step 6 did not always occur after step 5; it would sometimes occur immediately after step 1. If the e-prescription was subsequently changed, the EDS discharge medication list would not automatically update and instead needed to be reimported by clicking again on the link within the EDS).

After the EDS was signed-off by the hospital doctor, it was automatically transmitted electronically via a secure messaging service to the GP. If there was a transmission failure, the document was faxed or printed and posted.

Sample selection

A report of all patients discharged during the study period was generated. We sought to obtain a sample of approximately 100 patients from across the hospital, representative of adult admissions. The number of discharges included from each ward was based on the proportion of the hospital’s total discharges contributed by that ward during the study period. The sample size required from each ward was entered into Microsoft Excel to generate random number sequences to identify patients for inclusion. Additional random numbers were generated to replace patients who met exclusion criteria. Patients were excluded if they:

did not have a completed EDS in the medical record;

were admitted for less than 24 h;

were discharged to another hospital;

died in hospital;

did not take any medications prior to admission and were not prescribed medications on discharge;

were discharged from a slow-stream rehabilitation service (e.g. acquired brain injury unit);

had missing records that were required for the audit (e.g. pharmacist-verified “Medication History on Admission” form or pharmacist-reviewed and reconciled paper discharge prescription); and

had their discharge prescription prepared by a pharmacist as part of a coinciding pilot study that was occurring on two surgical wards to test a pharmacist-assisted discharge prescribing model (as this may have influenced the quality of medication information in the EDS).

Data collection

The audit tool was adapted from one used in an earlier study (Elliott et al., 2012). The tool was piloted on six EDS and modified to improve usability. The audit was conducted by two experienced pharmacists who were employees of the study hospital. They audited five EDS and PMLs together to ensure a consistent approach to data collection and interpretation and the remaining cases were audited by one pharmacist. If there was uncertainty about interpretation, it was discussed between the two pharmacists to reach agreement.

To assess the accuracy of medication lists in the EDS and the PML, these lists were compared with the scanned record of the pharmacist-reviewed, reconciled and dispensed paper discharge prescription (as this was considered to be the most accurate discharge medication list, reflecting exactly the medications the patient was intended to take after discharge).

Medication discrepancies were classified as (a) medication omission, (b) unintentional inclusion of medication, (c) strength or dose discrepancy or (d) strength or dose omission.

To assess the accuracy of medication change documentation in the EDS, the pharmacist-verified “Medication History on Admission” form was compared to the scanned record of the pharmacist-verified paper discharge prescription to identify changes to patients’ pre-admission medications that had been made in hospital. A medication change was defined as commencement of a new medication, cessation of a pre-admission medication or dose modification of a pre-admission medication. The EDS was then reviewed to determine whether these changes had been mentioned and whether the reason for the change was explained. If the reason for a change was not explicitly stated but was able to be easily deduced from the clinical synopsis or discharge diagnosis (e.g. commencement of a bisphosphonate for a patient who presented with a low-trauma bone fracture), it was considered to have been explained.

The hospital’s policy was that an EDS be prepared on transfer from one care type to another (e.g. emergency department to acute ward, or acute ward to subacute/rehabilitation unit), as well as on final discharge from the hospital. Therefore, some patients had more than one EDS for a single hospital admission. As all EDS are sent to patients’ GPs, all EDS were reviewed when determining whether medication changes were communicated and explained. Only the final EDS included a medication list because no discharge prescription is prepared for intra-hospital transfers.

Clinical significance of medication changes and EDS discrepancies

One investigator (YT) reviewed all medication changes and medication list discrepancies to identify those for which failure to accurately communicate in the EDS would be associated with a low risk of an adverse outcome. These were discussed with two other investigators, one a senior geriatrician (MD) and the other a senior pharmacist (RE), to reach expert consensus. Medication discrepancies and changes related to the following categories were agreed to be likely associated with a low risk of an adverse outcome: pro re nata (prn; “only if necessary”) aperients, prn simple analgesics (excluding opioids), prn lubricant eye drops, non-essential supplements (e.g. multivitamins), non-essential topical products (e.g. emollients), intermittent/infrequently administered medications that were not commenced or administered in hospital and were unrelated to the reason for admission (e.g. vitamin B12 three monthly injections in a surgical patient), and incorrect administration device recorded (provided the drug and dose were correct). For the purposes of this study, all other medication discrepancies and medication changes were classified as “potentially clinically significant.”

Primary outcome measures

proportion of the EDS with one or more medication list discrepancies;

median number of EDS medication list discrepancies per patient; and

proportion of medication changes that were stated and explained in the EDS.

Secondary outcome measures

proportion of the EDS that contained a discharge medication list; and

proportion of the PML with one or more medication list discrepancies.

The study was approved by the health service’s human research ethics committee.

Results

Study sample

A total of 144 patient discharges were reviewed, of which 87 were included in the audit (Table 1). Twenty-nine (20.1%) patients could not be included due to missing, incomplete or illegible records that were required for the audit, and 28 (19.4%) met other exclusion criteria: 16 (11.1%) patients were ineligible because they did not have a completed EDS, and 12 (8.3%) were ineligible because their discharge prescription was prepared by a pharmacist, they died in hospital or were transferred to another hospital (Table 2). Consideration was given to including additional records to meet our intended target of 100; however, the final sample of 87 was representative of the mix of hospital discharges over the study period so this was deemed unnecessary (Table 1).

Table 1.

Patient demographics (n = 87).

Age (years), median (IQR)	64 (47–77)
Gender
Males, n (%)	48 (55.2)
Females, n (%)	39 (44.8)
Median number of discharge medications (IQR)	6 (3–11)
Median number of medication changes (IQR)	4 (2–6)
Median number of potentially clinically significant medication changes (IQR)	3 (2–6)
Median length of hospital stay, days (IQR)	5 (3–10)
Ward type discharged from
Acute medical,^a n (%)	44 (50.6)
Acute surgical,^b n (%)	23 (26.4)
Subacute or rehabilitation,^c n (%)	12 (13.8)
Haematology, oncology or palliative care,^d n (%)	8 (9.2)

^aCardiology, endocrinology, epilepsy, gastroenterology, general medicine, infectious diseases, neurology, psychiatry (acute), renal, respiratory, spinal and urology (discharges from acute medical units comprised 51.8% of all hospital discharges over the study period).

^bCardiac; colorectal; ear, nose and throat; general; neuro; orthopaedic, plastic and vascular surgery (discharges from surgical units comprised 25.6% of all hospital discharges over the study period).

^cAged care, psychiatry (subacute), orthopaedic rehabilitation and neurology rehabilitation (discharges from subacute units comprised 13.3% of all hospital discharges over the study period).

^dDischarges from haematology, oncology or palliative care units comprised 9.3% of all hospital discharges over the study period.

Table 2.

Reasons for exclusion (n = 57).

Records required for the audit not available:	29
– No pharmacist-verified “Medication History on Admission” form (n = 13) – Scanned copy of the pharmacist-reviewed and reconciled discharge prescription missing or incomplete (n = 10)^a – Scanned copy of the pharmacist-reviewed and reconciled discharge prescription illegible (n = 3)^b – No hospital discharge prescription prepared by the medical team (n = 3)
No completed electronic discharge summary	16
Discharge prescription prepared by a pharmacist	10
Patient discharged to another hospital	1
Patient died during admission	1

^aFor six patients, the discharge prescription was not scanned after processing by the pharmacy department. For four patients, the scanned discharge prescription had one or more pages missing (discharge prescriptions are often spread over multiple printed pages, especially when there are late changes).

^bSecurity features on the Australian Government Pharmaceutical Benefits Scheme prescription paper (shading/watermark) resulted in scanned images of discharge prescriptions sometimes becoming illegible.

Accuracy of EDS medication lists

Eighty-five (98%) patients’ EDS included a list of discharge medications. Of these, 50 (59%) medication lists contained one or more discrepancies. A total of 131 medication list discrepancies were identified (median of one discrepancy per patient, inter-quartile range 1–2, range 0–15; Figure 1). Forty-seven discrepancies were considered to have low risk of leading to an adverse outcome, leaving 84 (64%) that were potentially clinically significant. Thirty-seven (44%) medication lists contained one or more potentially clinically significant discrepancies.

Figure 1.

Medication list discrepancies in electronic discharge summaries.

Communication of medication changes in the EDS

A total of 390 medication changes were made to patients’ pre-admission medications, of which 351 were considered to be potentially clinically significant. Of these, 162 (46%) were stated in the EDS and 153 (44%) were both stated and provided with a reason.

Accuracy of pharmacist-prepared PMLs

PMLs were available for 71 (81.6%) patients. Sixty-two (87%) of these were entirely consistent with the pharmacist-reviewed, reconciled and dispensed paper discharge prescription. A total of 15 medication discrepancies were identified in 9 PMLs, of which 11 were considered to be potentially clinically significant.

Discussion

There has been little previous research exploring the accuracy of medication lists in the EDS that are linked to an e-prescribing system. This study demonstrates that importing the discharge medication list from an electronic discharge prescription record does not guarantee concordance with the medications with which the patient is actually discharged. Although 98% of completed EDS included a medication list, discrepancies occurred in 59% of cases (median of one discrepancy per patient).

While several Australian studies have reported that discrepancies in the EDS are common, there is significant variability in reported discrepancy rates, ranging between 13% and 80% (Callen et al., 2008, 2010; Elliott et al., 2012; Tong et al., 2017). This is likely due to differences in the types of hospitals and patients studied, the processes by which discharge summaries were prepared and the methods used to identify and define discrepancies.

The number of medication discrepancies in the EDS in our study was lower than in an earlier study at the same hospital, conducted at a time when prescriptions were handwritten and EDS medication lists were manually typed by a hospital doctor (Elliott et al., 2012). At that time, 80% of EDS prepared for medical and aged care patients had one or more discrepancies, with a median of two discrepancies per patient (Elliott et al., 2012). This suggests that the integrated electronic system has led to some improvement compared to the stand-alone, manually typed EDS system.

Possible reasons for discrepancies in EDS medication lists

We believe that a major reason why medication list discrepancies continued to occur despite the use of an integrated system was the requirement in Australia that discharge medications are supplied pursuant to a paper prescription. The paper prescription is reviewed and reconciled by a pharmacist, and then used to dispense the medications. Prescribing errors may be corrected on the paper prescription but not necessarily corrected in the electronic record. As it is the electronic prescription record that populates the EDS, use of this hybrid paper–electronic prescribing system creates potential for discrepancies. The risk of medication errors associated with hybrid paper–electronic medication management systems has been recognised (Elliott et al., 2016; Sparnon, 2013), though not in the context of discharge prescriptions and EDS. Given the retrospective nature of our study, we were unable to determine the cause of each discrepancy and therefore the proportion that was likely to be associated with the hybrid system.

Another reason for medication list discrepancies may be the functionality of the EDS module and its integration with the electronic prescribing system. In the EDS, the medication list is a copy of the electronic discharge prescription record at the time it was imported into the EDS. If a prescriber makes subsequent changes to the electronic prescription, these changes are not automatically reflected in the EDS medication list and instead need to be manually reimported to update the EDS.

Possible reasons for incomplete explanation of medication changes in the EDS

Our study found that less than half of the changes made to patients’ pre-admission medications were mentioned or explained in the EDS, which is consistent with other Australian studies (Lehnbom et al., 2014; Tan et al., 2014). It is not surprising that poor documentation of medication changes continued to be common, since determining and entering this information into the EDS was a manual process. Determining what has changed can require a lengthy reconciliation process comparing the paper-documented pre-admission medication history with the electronic inpatient medication chart and discharge prescription. Documenting the reasons for changes is also a manual process, particularly as one would often need to look through the inpatient clinical record (progress notes) for this information, which can be difficult and time-consuming.

Possible strategies for improvement

In an environment that will continue to utilise a hybrid paper–electronic discharge medication prescribing system for the foreseeable future, our findings highlight an urgent need for new approaches to ensure medication information provided in the EDS is accurate, comprehensive and consistent with the medications and information provided to the patient being discharged.

One approach could be to establish a role for hospital pharmacists to review and contribute to the discharge medication information sections of the EDS. Pharmacists are well placed to perform this activity as they already routinely perform medication review and reconciliation on discharge to ensure the discharge prescription is accurate and to identify medication changes in order to communicate these to patients, carers and other healthcare providers (e.g. community pharmacies). The potential benefit of involving pharmacists is highlighted by the difference between the accuracy of the EDS medication list and the pharmacist-prepared PML in our study (41% vs. 87%). Tong et al. (2017) recently reported an absolute risk reduction of 46.5% for medication errors in discharge summaries when pharmacists completed the medication section, compared with doctor-prepared discharge summaries at an Australian hospital. Ooi et al. (2017) reported that a greater proportion of medication changes were documented in pharmacist-prepared discharge medication management summaries compared with doctor-prepared discharge summaries (72.8% vs. 31.5%). In a report on the outcomes of a Global Conference on the Future of Hospital Pharmacy (International Pharmaceutical Federation, 2009), 93% of attendants agreed that hospital pharmacists should “provide continuity of care by transferring patient medicines information as patients move between sectors of care.”

Lehnbom et al. (2014) discussed the use of mandatory fields to remind doctors to include pertinent information, which may encourage documentation of reasons for medication changes. However, success of such functions may be limited where information about medication changes needs to be manually extracted from the medical record. A complementary approach is to optimise functionality of e-prescribing systems to enable medication change information (including reasons for changes) to be captured prospectively in the electronic system, to either be auto-populated into the EDS or made available to assist the EDS author to more easily identify changes and the reasons for changes. The e-prescribing system should enable (or force) prescribers to record the reasons for changes at the time they make the change.

EDS medication list functionality should also be optimised, for example, to ensure that lists auto-update when changes are made to the electronic discharge prescription record after it has been imported into the EDS.

Strengths and limitations of the study

This study has strengths and limitations. Strengths include the fact that patients were randomly selected, and the sample was stratified to ensure it was representative of the typical mix of discharges from the hospital. The methodology acknowledged that medication discrepancies are likely to have varying levels of risk of adverse outcome, so low-risk discrepancies were identified, enabling the number of “potentially clinically significant” discrepancies and medication changes to be reported. A limitation was the high exclusion rate, most commonly due to missing or illegible records of pharmacist-verified medication histories and discharge prescriptions. It is possible that excluded patients had a different EDS discrepancy rate compared to our sample. Addressing this would be useful to improve the integrity of the hospital’s information record and enable more representative samples to be analysed in the future. Another limitation was the retrospective methodology, which meant we could not investigate the specific cause of each discrepancy. The use of pharmacists employed by the hospital to conduct the audit could potentially have introduced bias. Data collection by two pharmacists, without assessment of inter-rater reliability, was also a limitation. However, this was mitigated using a standardised data collection form and having the pharmacists audit five cases together to ensure a consistent approach to identifying medication changes and discrepancies. Finally, the method to determine “potentially clinically significant” medication discrepancies and changes was based only on the opinions of the investigators.

Conclusion

Discrepancies in medication information in EDSs continued to occur despite integration of the EDS with e-prescribing. Health services implementing electronic medication management systems and EDSs need to optimise the functionality and integration of electronic systems and implement risk mitigation strategies if a hybrid paper–electronic system is used. Pharmacists may have a role in ensuring medication information in the EDS is accurate.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the Clinical Leadership Group on Care of Older People in Hospital, Victorian Government Department of Health and Human Services.

References

Belleli

Naccarella

Pirotta

(2013) Communication at the interface between hospitals and primary care – a general practice audit of hospital discharge summaries. Australian Family Physician 42(12): 886–890.

Bergkvist

Midlov

Hoglund

. (2009) Improved quality in the hospital discharge summary reduces medication errors – LIMM: Landskrona Integrated Medicines Management. European Journal of Clinical Pharmacology 65(10): 1037–1046.

Callen

Alderton

McIntosh

(2008) Evaluation of electronic discharge summaries: a comparison of documentation in electronic and handwritten discharge summaries. International Journal of Medical Informatics 77(9): 613–620.

Callen

McIntosh

(2010) Accuracy of medication documentation in hospital discharge summaries: a retrospective analysis of medication transcription errors in manual and electronic discharge summaries. International Journal of Medical Informatics 79(1): 58–64.

Coleman

(2003) Falling through the cracks: challenges and opportunities for improving transitional care for persons with continuous complex care needs. Journal of the American Geriatrics Society 51(4): 549–555.

Cornu

Steurbaut

Leysen

. (2012) Discrepancies in medication information for the primary care physician and the geriatric patient at discharge. Annals of Pharmacotherapy 46(7–8): 983–990.

Elliott

Lee

Hussainy

(2016) Evaluation of a hybrid paper–electronic medication management system at a residential aged care facility. Australian Health Review 40(3): 244–250.

Elliott

Tran

Taylor

. (2012) Gaps in continuity of medication management during the transition from hospital to residential care: an observational study (MedGap Study). Australasian Journal on Ageing 31(4): 247–254.

Forster

Murff

Peterson

. (2003) The incidence and severity of adverse events affecting patients after discharge from the hospital. Annals of Internal Medicine 138(3): 161–167.

10.

Forster

Murff

Peterson

. (2005) Adverse drug events occurring following hospital discharge. Journal of General Internal Medicine 20(4): 317–323.

11.

Garrett

McCormack

(2014) Does an electronic discharge referral system improve the quality of medication prescribing? Journal of Pharmacy Practice and Research 44(1): 29–34.

12.

Hammad

Wright

Walton

. (2014) Adherence to UK national guidance for discharge information: an audit in primary care. British Journal of Clinical Pharmacology 78(6): 1453–1464.

13.

International Pharmaceutical Federation (2009) The Basel Statements on the future of hospital pharmacy. American Journal of Health System Pharmacy 66(Suppl 3): S61–S66.

14.

Kripalani

LeFevre

Phillips

. (2007) Deficits in communication and information transfer between hospital-based and primary care physicians: implications for patient safety and continuity of care. JAMA 297(8): 831–841.

15.

Lehnbom

Raban

Walter

. (2014) Do electronic discharge summaries contain more complete medication information? A retrospective analysis of paper versus electronic discharge summaries. Health Information Management Journal 43(3): 4–12.

16.

Mills

Weidmann

Stewart

(2016) Hospital discharge information communication and prescribing errors: a narrative literature overview. European Journal of Hospital Pharmacy 23(1): 3–10.

17.

Ooi

Rofe

Vienet

. (2017) Improving communication of medication changes using a pharmacist-prepared discharge medication management summary. International Journal of Clinical Pharmacy 39(2): 394–402.

18.

Roughead

Semple

Rosenfeld

(2016) The extent of medication errors and adverse drug reactions throughout the patient journey in acute care in Australia. International Journal of Evidence-Based Healthcare 14(3): 113–122.

19.

Shakib

Philpott

Clark

(2009) What we have here is a failure to communicate! Improving communication between tertiary to primary care for chronic heart failure patients. Internal Medicine Journal 39(9): 595–599.

20.

Sparnon

(2013) Spotlight on electronic health record errors: paper or electronic hybrid workflows. PA Patient Safety Advisory 10(2): 55–58.

21.

Stowasser

Collins

Stowasser

(2002) A randomised controlled trial of medication liaison services – patient outcomes. Journal of Pharmacy Practice and Research 32(2): 133–140.

22.

Tan

Mulo

Skinner

(2014) Transition from hospital to primary care: an audit of discharge summary – medication changes and follow-up expectations. Internal Medicine Journal 44(11): 1124–1127.

23.

Tong

Roman

Mitra

. (2017) Reducing medication errors in hospital discharge summaries: a randomised controlled trial. Medical Journal of Australia 206(1): 36–39.

24.

Witherington

Pirzada

Avery

(2008) Communication gaps and readmissions to hospital for patients aged 75 years and older: observational study. Quality and Safety in Health Care 17(1): 71–75.